Special crowd monitoring method based on LoRa wireless communication protocol

Abstract

The invention provides a special crowd monitoring method based on a LoRa wireless communication protocol, and relates to the technical field of Internet of Things positioning protection. According to the technical scheme provided by the invention, the initial risk score is finely corrected according to the influence factors to obtain the final risk score, and the target monitoring mode of the anti-lost terminal can be intelligently determined according to the score interval in which the final risk score is located. According to the differentiated monitoring strategy based on risk assessment, a high-frequency monitoring mode is adopted in a high-risk interval to ensure the monitoring precision, and a low-frequency monitoring mode is adopted in a low-risk interval to reduce the communication overhead, so that intelligent distribution and optimal configuration of monitoring resources are realized, the power consumption of equipment can be remarkably reduced on the premise of ensuring the monitoring precision, and the system reliability is improved. The equipment endurance time is prolonged, the charging frequency is reduced, and the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of IoT positioning and protection technology, specifically to a method for monitoring special populations based on the LoRa wireless communication protocol. Background Technology

[0002] With the deepening of societal aging and the increasing public awareness of family safety, the demand for monitoring technologies for special populations is becoming increasingly urgent. The special populations referred to in this application mainly include the elderly, children, and other individuals with relatively weak mobility and higher safety risks. These individuals are prone to getting lost, wandering off, or encountering accidents in daily life, requiring effective location monitoring and safety protection through technological means. Special population monitoring technology is widely applied in various scenarios such as home guardianship, community security, nursing home management, and kindergarten safety. Through real-time tracking and location services, it provides comprehensive safety guarantees for special populations, while simultaneously reducing the psychological burden on family members and guardians, and improving the overall level of social safety management.

[0003] Many technologies for preventing loss of individuals with special needs employ real-time monitoring systems based on GPS positioning and wireless communication. These systems track and monitor individuals by having them wear positioning devices that periodically report their location to a monitoring center. However, in practice, this approach suffers from a lack of intelligent judgment regarding the movement patterns and risk levels of the monitored individuals. This prevents the adjustment of communication strategies based on specific monitoring needs. To ensure monitoring accuracy, communication frequencies are often increased, employing high-frequency location reporting mechanisms to obtain more precise and timely location information. This leads to an inability to balance monitoring accuracy with device power consumption, resulting in rapid battery depletion and frequent charging or replacements, significantly reducing system usability and user experience. Summary of the Invention

[0004] This application provides a special population monitoring method based on the LoRa wireless communication protocol, which can reduce device power consumption and improve user experience while ensuring monitoring accuracy.

[0005] In a first aspect, this application provides a method for monitoring special populations based on the LoRa wireless communication protocol, the method comprising: The LoRa gateway receives the current status signal sent by the anti-loss terminal and calculates an initial risk score representing the current risk level based on the current status signal. The current status signal is input into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and to determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The initial risk score is revised based on the influencing factors to obtain the final risk score; Based on the score range of the final risk score, the target monitoring mode of the anti-loss terminal is determined so that the anti-loss terminal monitors the monitored object according to the target monitoring mode.

[0006] By adopting the above technical solution, after receiving the current status signal sent by the anti-loss terminal, an initial risk score is calculated based on this signal. The current status signal is then input into a historical behavior trajectory model trained based on the historical movement trajectory data of the monitored object, accurately predicting the monitored object's predicted route within a preset time period. Simultaneously, multiple influencing factors corresponding to each location point on the predicted route are determined. The initial risk score is refined based on these influencing factors to obtain the final risk score. The target monitoring mode of the anti-loss terminal can be intelligently determined based on the score range of the final risk score. This differentiated monitoring strategy based on risk assessment uses a high-frequency monitoring mode in high-risk areas to ensure monitoring accuracy, and a low-frequency monitoring mode in low-risk areas to reduce communication overhead. This achieves intelligent allocation and optimized configuration of monitoring resources, significantly reducing device power consumption, extending device battery life, reducing charging frequency, and improving user experience while ensuring monitoring accuracy.

[0007] Optionally, the current status signal includes the location information, motion status information, and environmental information of the anti-loss terminal, and the calculation of the initial risk score representing the current risk level based on the current status signal includes: Based on the location information, the safe distance between the current location of the anti-loss terminal and the boundary of the preset safe area is determined. The safe distance is compared with multiple preset distance thresholds, and the score corresponding to the distance interval to which the safe distance belongs is determined as the location risk score. Based on the motion state information, the moving speed of the anti-loss terminal is determined, the moving speed is compared with multiple preset speed thresholds, and the score corresponding to the speed range to which the moving speed belongs is determined as the motion risk score. Based on the environmental information, the environmental type of the anti-loss terminal is determined, and the corresponding risk coefficient is found in the preset environmental type-risk coefficient mapping table according to the environmental type. The risk coefficient is then determined as the environmental risk score. The location risk score, the movement risk score, and the environmental risk score are weighted and summed according to preset weighting coefficients to obtain the initial risk score.

[0008] Optionally, before inputting the current state signal into the historical behavior trajectory model, the method further includes: The monitored object acquires multiple historical movement trajectory data within a historical time period. Each historical movement trajectory data includes multiple historical location points arranged in chronological order and a timestamp corresponding to each historical location point. For each historical movement trajectory data, feature extraction is performed to obtain a trajectory feature vector, which includes movement direction feature, movement speed feature, dwell time feature, and trajectory repeatability feature; The trajectory feature vectors corresponding to the multiple historical movement trajectory data are used as training samples, and the training samples are trained using a sequence prediction model to obtain the historical behavior trajectory model.

[0009] Optionally, the step of inputting the current state signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period includes: The current state signal is input into the historical behavior trajectory model to extract the current trajectory feature vector, which includes the current movement direction feature, the current movement speed feature, and the current position feature. Based on the current trajectory feature vector, sequence prediction is performed using the historical behavior trajectory model to generate multiple predicted location points within the preset time period. The multiple predicted location points are then connected in chronological order to form the predicted route.

[0010] Optionally, determining the multiple influencing factors corresponding to each location point on the predicted route includes: For each predicted location point on the predicted route, a preset regional attribute database is queried based on the coordinate information of the predicted location point to obtain the regional risk level corresponding to each predicted location point. For adjacent predicted location points on the predicted route, the ratio of the distance between adjacent predicted location points to the time interval is calculated to obtain the predicted movement speed corresponding to the predicted location point. For each predicted location point on the predicted route, a preset environment database is queried based on the coordinate information of the predicted location point to obtain the predicted environment type corresponding to the predicted location point; The regional risk level, the predicted movement speed, and the predicted environment type are considered as multiple influencing factors corresponding to the predicted location point.

[0011] Optionally, the step of revising the initial risk score based on the influencing factors to obtain the final risk score includes: Each of the aforementioned influencing factors is compared with a preset regional correction table, a preset speed correction table, and a preset environmental correction table to obtain the correction value corresponding to each of the aforementioned influencing factors; The initial risk score is corrected based on the aforementioned correction values ​​to obtain the final risk score.

[0012] Optionally, the method further includes: Based on the historical movement trajectory data of the monitored object, trajectories that repeat more than a preset threshold number of times are identified as fixed travel routes. Based on the average historical risk scores of each location point on the fixed travel route, a historical risk score curve corresponding to the fixed travel route is constructed. Calculate the gradient of the score change in the historical risk score curve, and determine the points where the gradient of the score change is greater than a preset gradient threshold as feedback monitoring nodes; When the overlap between the predicted route and the fixed travel route is greater than a preset overlap threshold, it is determined that the monitored object is moving along the fixed travel route. The anti-loss terminal is controlled to report the actual status signal to the LoRa gateway only when it reaches the feedback monitoring node, and the reporting frequency is reduced when it is not at the feedback monitoring node.

[0013] Secondly, this application provides a special population monitoring system based on the LoRa wireless communication protocol, the system comprising: The initial scoring module is used to receive the current status signal sent by the anti-loss terminal through the LoRa gateway, and calculate an initial risk score representing the current risk level based on the current status signal. The prediction module is used to input the current status signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and to determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The correction module is used to correct the initial risk score based on the influencing factors to obtain the final risk score; The monitoring module is used to determine the target monitoring mode of the anti-loss terminal based on the score range of the final risk score, so that the anti-loss terminal monitors the monitored object according to the target monitoring mode.

[0014] Thirdly, this application provides a computer storage medium storing a plurality of instructions adapted for loading by a processor and executing any of the methods described above.

[0015] Fourthly, this application provides an electronic device including a processor, a memory, and a transceiver, wherein the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform any of the methods described above.

[0016] In summary, the beneficial effects of the technical solution of this application include: By adopting the above technical solution, after receiving the current status signal sent by the anti-loss terminal, an initial risk score is calculated based on this signal. The current status signal is then input into a historical behavior trajectory model trained based on the historical movement trajectory data of the monitored object, accurately predicting the monitored object's predicted route within a preset time period. Simultaneously, multiple influencing factors corresponding to each location point on the predicted route are determined. The initial risk score is refined based on these influencing factors to obtain the final risk score. The target monitoring mode of the anti-loss terminal can be intelligently determined based on the score range of the final risk score. This differentiated monitoring strategy based on risk assessment uses a high-frequency monitoring mode in high-risk areas to ensure monitoring accuracy, and a low-frequency monitoring mode in low-risk areas to reduce communication overhead. This achieves intelligent allocation and optimized configuration of monitoring resources, significantly reducing device power consumption, extending device battery life, reducing charging frequency, and improving user experience while ensuring monitoring accuracy. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a special population monitoring method based on the LoRa wireless communication protocol according to an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a special population monitoring system based on the LoRa wireless communication protocol according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0018] Explanation of reference numerals in the attached drawings: 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0020] In the description of the embodiments of this application, words such as "illustrative," "for example," or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "illustrative," "for example," or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Rather, the use of words such as "illustrative," "for example," or "for example" is intended to present the relevant concepts in a specific manner.

[0021] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0022] Please see Figure 1 This is a flowchart illustrating a method for monitoring special populations based on the LoRa wireless communication protocol, provided in an embodiment of this application. This method can be implemented using a computer program, a microcontroller, or run on a special population monitoring system based on the LoRa wireless communication protocol and the von Neumann architecture. The computer program can be integrated into the application or run as a standalone utility application. The specific steps of the method for monitoring special populations based on the LoRa wireless communication protocol are described in detail below.

[0023] S101: Receive the current status signal sent by the anti-loss terminal through the LoRa gateway, and calculate the initial risk score representing the current risk level based on the current status signal; LoRa gateways, based on the LoRa long-range wireless communication protocol, are network access devices used to receive and forward data signals from anti-loss terminals. They feature low power consumption, long-distance transmission, and strong penetration, achieving an effective communication range of 3-5 kilometers in urban environments. Anti-loss terminals are portable positioning devices worn on the monitored object, integrating hardware components such as GPS positioning modules, motion sensors, and environmental sensors. They can collect and transmit location and movement status information in real time. The current status signal represents the comprehensive status data packet collected by the anti-loss terminal at a specific moment, including multi-dimensional information such as location coordinates, movement speed, direction of movement, acceleration, ambient temperature and humidity, and light intensity. The initial risk score is a value calculated based on the current status signal using a preset algorithm, used to quantify the current security risk level of the monitored object.

[0024] Specifically, the system first establishes a communication connection with the anti-loss terminal through a LoRa gateway. After receiving the data packets sent by the anti-loss terminal, the gateway performs protocol parsing and data verification, extracting the current status signal containing location information, motion status information, and environmental information. Subsequently, the system performs multi-dimensional analysis of the current status signal, calculating an initial risk score based on indicators such as position deviation, abnormal movement speed, and environmental hazard. During the calculation process, the system compares the location information with the preset safe zone boundary, calculates the safe distance, and maps it to a location risk score; it analyzes the speed data in the motion status information to determine whether there are abnormally rapid movements or prolonged periods of stillness, calculating a motion risk score; and it assesses the safety of the environment based on environmental information, such as whether it is in a high-risk environment like a busy traffic area, construction site, or near water, calculating an environmental risk score. Finally, the various risk scores are weighted and summed according to preset weights to obtain a comprehensive initial risk score.

[0025] S102: Input the current status signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The historical behavior trajectory model represents a behavior prediction model built based on machine learning algorithms. It learns the behavioral patterns and movement rules of the monitored object by analyzing its historical movement data, enabling it to predict the behavior trajectory within a certain future timeframe. This model typically employs deep learning architectures such as recurrent neural networks, long short-term memory networks, or attention mechanisms. For example, the LSTM-RNN model can effectively capture long-term dependencies in time-series data. The preset time period refers to the future prediction time range set by the system, usually determined based on the application scenario and monitoring requirements, such as 15 minutes, 30 minutes, or 1 hour. The predicted route represents the sequence of paths the monitored object may traverse within the preset time period, consisting of multiple predicted location points arranged in chronological order. Each location point includes latitude and longitude coordinates and a corresponding timestamp. Influencing factors refer to various environmental and behavioral elements that affect the risk score calculation, including regional risk level, predicted movement speed, and predicted environment type.

[0026] Specifically, the system first inputs feature data such as location information, direction of movement, and speed from the current state signal into a pre-trained historical behavior trajectory model. Based on the current state features and learned historical behavior patterns, the model generates a sequence of location points within a preset future time period using a sequence prediction algorithm. During the prediction process, the model considers factors such as the monitored object's personal behavioral habits, common routes, and activity time patterns, combined with external conditions such as the current time, date, and weather, outputting a predicted route containing timestamps and coordinate information. Subsequently, the system performs a detailed analysis of each location point on the predicted route. It obtains the regional risk level of each location point by querying a preset regional attribute database, determines the predicted movement speed by calculating the distance and time interval between adjacent location points, and obtains the predicted environmental type information by querying an environmental database, comprehensively determining multiple influencing factors corresponding to each location point.

[0027] In some embodiments, trajectory prediction and influencing factor determination can be achieved in multiple ways: Optionally, a deep learning prediction method is adopted. The system converts the current state signal into a feature vector format, including normalization of position coordinates, vectorization of movement speed, and periodic encoding of time features, and then inputs it into a historical behavior trajectory model based on an LSTM architecture. The model analyzes the similarity between the current state and historical patterns through a multi-layer neural network structure, identifies key behavioral features using an attention mechanism, generates a sequence of future location coordinates through a sequence-to-sequence prediction method, and then performs spatial analysis on the predicted route. It uses the buffer analysis function of the geographic information system to determine points of interest and hazards around the route, calculates the comprehensive risk index of each location point through spatial overlay analysis, and finally forms predicted route data containing detailed influencing factors.

[0028] S103: The initial risk score is revised based on the influencing factors to obtain the final risk score; The influencing factors refer to various elements identified in the predicted route analysis that may affect the safety status of the monitored object, including spatial and behavioral attributes such as regional risk level, predicted movement speed, and predicted environment type. The final risk score represents the revised comprehensive risk assessment result, which is more accurate and comprehensive than the initial risk score and can better reflect the overall safety risk level of the monitored object within the preset time period.

[0029] Specifically, the system first classifies and quantifies the various influencing factors identified along the predicted route, converting qualitative factors into calculable numerical values. For regional risk levels, the system maps different risk levels to corresponding correction values ​​using a pre-set regional correction table; for example, a high-risk area corresponds to a correction value of +15 points, and a low-risk area corresponds to a correction value of -5 points. For predicted movement speed, the system converts abnormal movement speeds into risk correction values ​​using a speed correction table; for example, excessively fast movement may increase the risk score by 10 points, and excessively slow movement may increase it by 5 points. For predicted environment type, the system uses an environment correction table to determine the risk adjustment amount under different environmental conditions; for example, rainy weather increases the risk by 8 points, and nighttime weather increases it by 12 points. After obtaining the correction values ​​corresponding to all influencing factors, the system adjusts the initial risk score using a weighted average or direct summation method, while considering the confidence level and time decay factor of the correction values ​​to ensure the accuracy and reasonableness of the correction results, ultimately outputting an optimized final risk score.

[0030] In some embodiments, the risk score correction calculation can be implemented in multiple ways: Optionally, a linear weighted correction method is adopted. The system first sorts all influencing factors on the predicted route by importance, assigns different weight coefficients according to the factor type and degree of influence, then precisely matches each influencing factor with the corresponding correction table to obtain the basic correction value of each factor. Next, considering the spatial distribution characteristics and time delay effect of the factors, the basic correction value is processed by distance decay and time decay to calculate the adjusted actual correction value. Finally, all correction values ​​are weighted and summed according to the preset weight coefficients, and the result is linearly combined with the initial risk score to obtain the final risk score result. Boundary value checks are performed on the result to ensure that the score is within the effective range.

[0031] S104: Based on the score range of the final risk score, determine the target monitoring mode of the anti-loss terminal so that the anti-loss terminal monitors the monitored object in accordance with the target monitoring mode.

[0032] The final risk score represents a comprehensive risk assessment value after adjustments for influencing factors, more accurately reflecting the current and future safety risk status of the monitored object. The score range refers to dividing the risk score range into several consecutive numerical segments, each corresponding to a different risk level. The target monitoring mode represents the equipment operating strategy determined based on the risk level, including parameter configurations such as data reporting frequency, positioning accuracy requirements, sensor sampling intervals, and power management strategies. Different risk levels correspond to different monitoring intensities.

[0033] Specifically, the system first determines the risk level range based on the final risk score, then queries the preset monitoring mode configuration table to obtain the target monitoring mode parameters corresponding to that risk level. In low-risk situations, the system configures a lower reporting frequency, such as sending location information every 15-30 minutes, using standard-precision GPS positioning, and enabling low-power mode to extend device battery life. In medium-risk situations, the system increases the reporting frequency to once every 5-10 minutes, increases the sampling frequency of the motion sensor, and enables environmental monitoring. In high-risk situations, the system switches to a high-frequency monitoring mode, reporting detailed status information every 1-3 minutes, enabling high-precision positioning, and activating emergency call and alarm functions. In extremely high-risk situations, the system enters real-time monitoring mode, continuously sending location information and status data, while simultaneously sending emergency notifications to the guardian. The system also sends the target monitoring mode configuration instructions to the anti-loss terminal via the LoRa gateway. Upon receiving the instructions, the terminal updates its local operating parameters and executes the location information collection and transmission tasks according to the new monitoring mode, achieving risk-adaptive intelligent monitoring.

[0034] In some embodiments, the determination and configuration of the monitoring mode can be achieved in multiple ways: Optionally, a hierarchical threshold judgment method can be adopted. The system presets multiple risk score thresholds as interval demarcation points, such as setting 20, 50, and 80 as demarcation thresholds, dividing the score range of 0-100 into four level intervals. After the final risk score is calculated, the system determines the specific interval to which the score belongs through simple numerical comparison. Then, according to the interval level, it queries the preset monitoring mode parameter table to obtain a complete set of parameters, including reporting interval, positioning accuracy, sensor configuration, power consumption strategy, etc. Then, the system generates a standardized configuration instruction data packet, which is sent to the corresponding anti-loss terminal through the downlink channel of the LoRa gateway. After receiving and parsing the instruction, the terminal updates its local configuration and begins to perform data collection and transmission tasks according to the new monitoring mode. At the same time, it sends back the configuration update confirmation information to the monitoring center.

[0035] Based on the above embodiments, as an optional implementation method, the current status signal includes the location information, motion status information and environmental information of the anti-loss terminal. The method of calculating the initial risk score representing the current risk level based on the current status signal in S101 can be implemented through the following steps S201-S204.

[0036] In this application, the preset security zone is implemented using electronic fence technology. The electronic fence adopts a customizable security zone setting method, including a circular fence centered on the gateway and a polygonal fence based on geographical boundaries. The setting of the electronic fence is based on three stages: data acquisition, algorithm generation, and dynamic adjustment. In the data acquisition stage, historical trajectory data, time rule data, and static geographic data for 7-14 days are acquired to provide a basis for fence delineation. In the algorithm generation stage, a density clustering algorithm is used to aggregate high-frequency trajectory points to generate core activity areas. Combined with static geographic boundaries, the fence shape is adjusted, and multi-level rules for dividing security zones, warning zones, and danger zones are automatically defined. In the dynamic adjustment stage, the fence is adaptively updated based on real-time trajectory correction, temporary scene triggering, and device status linkage.

[0037] In addition to location, motion, and environmental information, the current status signal also includes device status information collected by a 9-axis motion sensor, used to detect abnormal states such as falls and continuous stillness. When the anti-loss terminal is within a preset safe area, it uses a low-power mode to upload only the RSSI value. When it exceeds the fence range, an alarm is immediately triggered and complete status information is uploaded.

[0038] S201: Determine the safe distance between the current location of the anti-loss terminal and the boundary of the preset safe area based on location information, compare the safe distance with multiple preset distance thresholds, and determine the score corresponding to the distance interval to which the safe distance belongs as the location risk score; The preset safe zone boundary refers to the geographical limit of the monitored object's permitted activity range set during system initialization, typically using geometric shapes such as polygons or circles to define the specific spatial range. The safe distance represents the straight-line distance from the current location of the anti-loss terminal to the nearest safe zone boundary line; this distance value reflects the degree to which the monitored object deviates from the safe range. The distance threshold represents multiple pre-set distance demarcation points used to divide the safe distance into different risk level intervals. The location risk score refers to the numerical risk assessment result assigned based on the safe distance interval.

[0039] Specifically, the preset safe zone boundaries are defined through electronic fence boundaries. The boundary shapes include circular fences with a radius of 100 meters centered on the gateway, and polygonal fences based on the geographical boundaries of closed areas such as residential communities and schools. The calculation of safe distances not only considers the straight-line distance to the nearest boundary, but also incorporates the multi-level zone settings of the fences, dividing the safe distance into the safe zone distance, the warning zone distance, and the danger zone distance.

[0040] The specific steps involved in constructing a pre-defined safe zone using electronic fence technology include: Data collection phase: Obtain historical trajectory data of the monitored object for 7-14 days, identify high-frequency areas as core activity areas; obtain time rule data, associate time periods with activity ranges; connect to map API to obtain the geographical boundaries of closed areas and the coordinates of dangerous areas; Algorithm generation stage: Density clustering algorithm is used to aggregate high-frequency trajectory points to generate core activity areas; fence shape is adjusted in combination with static geographic boundaries; multi-level rule settings are automatically set to divide safe zones, warning zones, and danger zones; Dynamic adjustment phase: Automatically expand the fence range based on new high-frequency areas appearing for three consecutive days; automatically generate a temporary fence when detecting movement to an unfamiliar location; temporarily adjust the fence warning range based on abnormal behavior detected by the device.

[0041] Specifically, firstly, latitude and longitude coordinates are extracted from the location information reported by the anti-loss terminal. Then, a geographic information processing algorithm is used to calculate the shortest distance between this coordinate point and the boundary of a preset safe zone. The calculation employs a point-to-polygon distance algorithm, traversing all line segments of the safe zone boundary and calculating the perpendicular distance from the current location point to each boundary line segment. The minimum value is taken as the safe distance. When the location point is inside the safe zone, the safe distance is negative or zero; when the location point is outside the safe zone, the safe distance is positive. After obtaining the safe distance, it is compared one by one with a preset distance threshold sequence to determine the specific range within which the safe distance falls. The distance thresholds are typically set as an increasing sequence, with each range corresponding to a different risk level and a corresponding score. The greater the distance, the more serious the deviation from the safe zone, and the higher the corresponding location risk score. Finally, based on the range to which the safe distance belongs, the corresponding location risk score is retrieved from a preset range score mapping table.

[0042] S202: Determine the moving speed of the anti-loss terminal based on motion status information, compare the moving speed with multiple preset speed thresholds, and determine the score corresponding to the speed range to which the moving speed belongs as the motion risk score; Motion status information refers to the dynamic data collected by the anti-loss terminal through its built-in motion sensors, including multi-dimensional motion characteristic parameters such as three-axis acceleration, angular velocity, and direction of movement. Movement speed represents the displacement of the monitored object per unit time, calculated by analyzing continuous changes in position coordinates or motion sensor data. Speed ​​threshold represents a pre-set speed boundary value based on the monitored object's behavioral characteristics and safety requirements, used to distinguish between different motion states such as normal movement, abnormally fast movement, and abnormally slow movement. Motion risk score represents the numerical result of the degree of risk of the movement behavior assessed based on the movement speed.

[0043] Specifically, the current movement speed is determined by analyzing the motion status information reported by the anti-loss terminal. The calculation methods include speed calculation based on location changes and speed estimation based on sensor data. The location-change-based method calculates the average movement speed by comparing the coordinate difference and time interval between two consecutive location reports. The sensor-data-based method estimates the real-time movement speed by analyzing the output of the accelerometer, combining integration and filtering. After obtaining the movement speed, it is compared and analyzed with a preset speed threshold sequence. Speed ​​thresholds typically include several boundary points such as a stationary threshold, a normal walking speed range, a fast movement threshold, and an abnormally high speed threshold. By comparing, the speed range to which the current movement speed belongs is determined; different ranges represent different levels of movement risk. Stationary or extremely slow movement indicates that the monitored object may be encountering difficulties or is in a dangerous state, while abnormally fast movement indicates a possible emergency or unexpected event. Based on the movement speed range, the corresponding movement risk score is obtained from a preset speed risk score table.

[0044] S203: Determine the environmental type of the anti-loss terminal based on environmental information, find the corresponding risk coefficient in the preset environmental type-risk coefficient mapping table according to the environmental type, and determine the risk coefficient as the environmental risk score. Among them, environmental information refers to the surrounding environmental characteristic data collected by the anti-loss terminal through environmental sensors, including physical environmental parameters such as temperature, humidity, light intensity, noise level, and air pressure. Environmental type represents the classification of the environment derived from comprehensive analysis of environmental information, reflecting the specific environmental conditions of the monitored object. The environmental type-risk coefficient mapping table is a pre-established table corresponding to environmental classifications and risk assessment coefficients, recording the standardized risk coefficient values ​​corresponding to different environmental types. The environmental risk score is used to represent the quantitative result of the degree of environmental-related risk assessed based on the environmental type.

[0045] Specifically, multi-sensor data fusion technology is used to comprehensively analyze various environmental parameters. The environmental type identification process employs a combination of rule-based reasoning and pattern matching, analyzing sensor data through pre-defined environmental discrimination rules. Light intensity is used to determine indoor and outdoor environments and day / night conditions; temperature and humidity data are used to identify special climatic conditions; noise levels are used to determine traffic conditions and population density; and air pressure changes are used to detect altitude changes and weather changes. By comprehensively analyzing these environmental parameters, the specific environmental type is determined, such as a quiet indoor environment, a normal outdoor environment, a busy traffic area, an adverse weather environment, or a nighttime environment. After determining the environmental type, the corresponding risk coefficient is looked up in a pre-defined environmental type-risk coefficient mapping table. The mapping table records standardized risk assessment values ​​for various environmental types, which are determined based on historical security event statistics and expert evaluations. Different environmental types pose different levels of security threats to the monitored objects, and their corresponding risk coefficients also differ. Once the corresponding risk coefficient is found, that value is directly assigned as the environmental risk score.

[0046] S204: The location risk score, movement risk score, and environmental risk score are weighted and summed according to preset weighting coefficients to obtain the initial risk score.

[0047] Specifically, the location risk score, movement risk score, and environmental risk score calculated in the aforementioned three steps are used as input parameters and weighted and summed according to preset weighting coefficients. The weighting coefficients are set based on the degree of impact of different risk factors on the overall safety situation, and the optimal weighting configuration is usually determined through historical data analysis, expert evaluation, and feedback from practical applications. Location risk typically carries a high weight because deviation from safe areas is the most direct indicator of safety threats. Movement risk has a lower weight, as abnormal movement patterns often foreshadow potential dangers. Environmental risk has a relatively low weight, but it still plays an important role under certain conditions. The weighted summation calculation process first multiplies each risk score by its corresponding weighting coefficient to obtain a weighted score. Then, the three weighted scores are summed to obtain the comprehensive risk score. To ensure the reasonableness of the calculation results, boundary checks and normalization are performed on the final result to keep the initial risk score within a preset numerical range.

[0048] Based on the above embodiments, as an optional implementation method, the construction of the historical behavior trajectory model before S102 can be specifically achieved through the following steps S301-S303.

[0049] S301: Obtain multiple historical movement trajectory data of the monitored object within a historical time period. Each historical movement trajectory data includes multiple historical location points arranged in chronological order and the timestamp corresponding to each historical location point. Historical time period refers to a specific time interval that the system traces backward, used to collect data on the past activities of the monitored object. Historical movement trajectory data represents the complete movement path record of the monitored object within a certain period of time, stored in digital form as a sequence of spatial location changes. Historical location points refer to the specific geographic coordinate information recorded in the trajectory data, including longitude and latitude values. Timestamps are used to identify the precise time each location point was recorded, stored in a standard time format.

[0050] Specifically, the system retrieves historical trajectory records of the monitored object by accessing the system database or storage service. The data acquisition process first determines the start and end times of the historical time period, typically set to a range of 7 to 30 days to ensure sufficient sample data for model training. The system then filters all location records within the specified time period from the trajectory data storage table and sorts them according to the data collection time. Each historical movement trajectory represents a complete travel process, a continuous sequence of location points from the starting position to the ending position. Data extraction requires integrity verification to ensure that each location point contains valid latitude and longitude coordinates and corresponding timestamp information. For location points with missing or abnormal data, interpolation methods are used for repair or direct removal. The acquired trajectory data is organized into an ordered sequence according to chronological order, providing standardized input data for subsequent feature extraction and model training. The system also needs to preprocess the trajectory data, including coordinate system unification, time format standardization, and duplicate data removal, to ensure that the data quality meets the analysis requirements.

[0051] S302: Extract features from each historical movement trajectory data to obtain a trajectory feature vector. The trajectory feature vector includes movement direction features, movement speed features, dwell time features, and trajectory repeatability features. Movement direction features represent the azimuth information between adjacent points on the trajectory, reflecting the movement trend and path selection preferences of the monitored object. Movement speed features refer to the statistical distribution of speed across different segments of the trajectory, including numerical features such as average speed, maximum speed, and speed variation. Dwell time features describe the distribution of the monitored object's dwell time at specific locations or areas, reflecting activity patterns and points of interest. Trajectory repeatability features represent the similarity between the current trajectory and historical trajectories, quantifying behavioral regularity and predictability. The trajectory feature vector is a numerical array formed by combining the above multi-dimensional features, used as input for machine learning algorithms.

[0052] Specifically, multi-dimensional feature extraction is performed on each historical trajectory data. The calculation of movement direction features involves analyzing the coordinate differences between adjacent points, calculating the azimuth using the arctangent function, and statistically analyzing the main movement direction, frequency of direction changes, and distribution of turning angles for the entire trajectory. Movement speed features are extracted based on the distance and time interval between adjacent points, calculating the instantaneous speed sequence, and then statistically analyzing descriptive features such as the mean, variance, and quantiles of the speed. Dwell time features are achieved by identifying dwell points in the trajectory; when multiple consecutive points cluster within a small area and the duration exceeds a threshold, it is marked as a dwelling behavior, and the number of dwellings, dwell time distribution, and dwell location features are statistically analyzed. Trajectory repetition features employ trajectory similarity calculation methods, comparing the current trajectory with historical trajectories to calculate spatial overlap, temporal similarity, and pattern matching. The feature extraction process uses a sliding window technique to handle long trajectories, dividing the trajectory into fixed-length segments and calculating the feature values ​​for each segment. All feature values ​​are normalized to eliminate dimensional differences between different features. Finally, all features are combined in a predetermined order into a fixed-dimensional feature vector, which serves as the standard input format for the machine learning model.

[0053] S303: Use the trajectory feature vectors corresponding to multiple historical movement trajectory data as training samples, and use a sequence prediction model to train the training samples to obtain a historical behavior trajectory model.

[0054] Sequence prediction models refer to machine learning algorithms specifically designed for processing time series data. They can learn the temporal dependencies and patterns in a sequence to predict future trends. Historical behavior trajectory models represent computational models trained to predict the movement behavior of monitored objects, including learned parameters and inference logic.

[0055] Specifically, the extracted trajectory feature vectors are organized into a machine learning training dataset, and the model training process is executed. The training dataset is constructed using a sliding time window method, taking a continuous sequence of feature vectors as input and the corresponding next time step or the next trajectory segment as the prediction target. The dataset is divided into training, validation, and test sets, typically allocated in a 70%, 15%, and 15% ratio to ensure the effectiveness and generalization ability of the model training. The sequence prediction model uses a recurrent neural network or long short-term memory network architecture, which has the ability to handle variable-length sequences and learn long-term dependencies. The model training process includes four main steps: forward propagation, loss calculation, backpropagation, and parameter update. In the forward propagation stage, the feature vectors are input into the network, and the predicted output is obtained through multi-layered neurons. In the loss calculation stage, the difference between the predicted result and the true label is compared, and the prediction error is quantified using loss functions such as mean squared error or cross-entropy. In the backpropagation stage, the gradient of the loss function with respect to the parameters of each layer is calculated to determine the direction of parameter adjustment. In the parameter update stage, the gradient descent algorithm is used to optimize the network parameters and reduce the prediction error. The training process employs batch processing and iterative optimization strategies, gradually improving model performance through multiple rounds of training. After training, the model is validated and tested to evaluate prediction accuracy and generalization performance, ensuring that the model achieves the expected prediction results.

[0056] Based on the above embodiments, as an optional implementation method, the current status signal is input into the historical behavior trajectory model in S102 to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period. This can be achieved through the following steps.

[0057] Input the current state signal into the historical behavior trajectory model and extract the current trajectory feature vector. The current trajectory feature vector includes the current movement direction feature, the current movement speed feature, and the current position feature. The current trajectory feature vector refers to a set of numerical features extracted from real-time status signals that describe the current movement state of the monitored object, used as input to the prediction model for analysis. The current movement direction feature represents the angle and trend of the monitored object's current movement direction, derived by analyzing the coordinate changes of the most recent few locations. The current movement speed feature refers to the monitored object's current movement speed value and speed change pattern, reflecting its current movement intensity and activity level. The current location feature describes the geographic coordinates of the monitored object's current location and its spatial relationship with historical activity areas.

[0058] Specifically, the current status signal obtained from the anti-loss terminal is first passed as input data to the trained historical behavior trajectory model. After receiving the current status signal, the model starts the feature extraction module to process the real-time data. The current movement direction feature is extracted by analyzing the coordinate relationship between the current position and the previous few historical position points, calculating the azimuth angle between adjacent position points, and identifying the main movement direction and the frequency of direction change. The system uses a sliding window method to select the 5 to 10 most recent position points for direction analysis to ensure the accuracy and timeliness of the direction features. The current movement speed feature is calculated based on the change in position coordinates and the corresponding time interval. Instantaneous speed is obtained through distance calculation and time difference, while analyzing the speed change trend and fluctuation. The current position feature includes not only latitude and longitude coordinate information, but also the distance relationship between the current position and historical high-frequency activity areas, the type of environment, and the relative position relationship with the preset safety area. The feature extraction process uses the same data processing method as the training phase to ensure the consistency of the feature format. All extracted feature values ​​are standardized and converted to the same numerical range and format as the training data. Finally, the various features are combined in a predetermined order to form the current trajectory feature vector, which serves as the input data for the model prediction stage.

[0059] Based on the current trajectory feature vector, sequence prediction is performed using a historical behavior trajectory model to generate multiple predicted location points within a preset time period. These multiple predicted location points are then connected in chronological order to form a predicted route.

[0060] Sequence prediction refers to the computational process of using a trained model to predict the future trend of a time series based on current input data, generating an estimate of the future state through the model's reasoning ability. The preset time period represents the future time range for trajectory prediction, typically set to a time window of 10 minutes to 1 hour. The predicted route is a complete movement path formed by connecting multiple predicted location points in chronological order, showcasing the expected future movement trajectory of the monitored object.

[0061] Specifically, the current trajectory feature vector is input into the historical behavior trajectory model to perform sequence prediction calculations. The model first encodes the input feature vector and extracts deep features and pattern information through the hidden layers of the neural network. The prediction process uses a recursive calculation method. The model calculates the position prediction result for the next moment based on the current input and internal state, and then uses the prediction result as new input to continue predicting the position for subsequent moments. This recursive prediction method continues until the entire preset time period is covered. During the prediction calculation process, the model uses the behavioral patterns and regularities learned in the training phase, combined with the current movement state and historical trajectory features, to infer the movement intention and target direction of the monitored object. Each prediction step generates a predicted position point containing latitude and longitude coordinates and assigns a corresponding timestamp to identify the prediction time. To improve prediction accuracy, the system uses an ensemble prediction method, running multiple predictions and performing statistical analysis on the results to select the most representative prediction path. The prediction process also includes uncertainty assessment, calculating the confidence interval for each predicted position point to indicate the reliability of the prediction result. The generated multiple predicted position points are arranged in timestamp order, and adjacent position points are connected by linear interpolation or curve fitting methods to form a continuous prediction route. The predicted route also needs to be matched and corrected with map data to ensure that the predicted path conforms to the actual road network and geographical constraints.

[0062] Based on the above embodiments, as an optional implementation method, the method of determining multiple influencing factors corresponding to each location point on the prediction route in S102 can be specifically implemented through the following steps S401-S404.

[0063] S401: For each predicted location point on the predicted route, query the preset regional attribute database based on the coordinate information of the predicted location point to obtain the regional risk level corresponding to each predicted location point; The regional attribute database refers to a pre-established data storage system containing security risk classification information for various geographical regions. It categorizes and manages different regions according to their geographical coordinate ranges, assigning risk levels accordingly. Regional risk levels indicate the degree of security risk in a specific geographical area, typically categorized as low, medium, or high risk, reflecting the probability and severity of security incidents occurring in that region.

[0064] Specifically, the system performs a regional risk level query operation for each predicted location point along the predicted route. First, the system extracts the latitude and longitude coordinates of each predicted location point as the retrieval criteria for the database query. The regional attribute database uses a spatial index structure to store risk level data, dividing the geographic space into grid cells or polygonal regions, with each spatial cell associated with a corresponding risk level identifier. The query process determines the spatial cell to which the predicted location point belongs through spatial geometric calculations, and uses a point-within-a-polygon judgment algorithm or a nearest neighbor search algorithm to achieve accurate spatial matching. For location points located at the boundaries of multiple risk areas, the system uses a distance-weighted method to comprehensively calculate the risk level, selecting the risk level of the nearest region as the attribute value for that point. Database queries employ batch processing to improve query efficiency, combining the coordinate information of multiple predicted location points into a query request, and obtaining the risk level information of all location points through a single database access. The query results include a numerical risk level corresponding to each predicted location point: low-risk areas are assigned a value of 1, medium-risk areas are assigned 2, and high-risk areas are assigned 3. The system also needs to handle query anomalies; for areas not covered in the database, the system supplements them with the default risk level or the risk level of neighboring areas.

[0065] S402: For adjacent predicted location points on the predicted route, calculate the ratio of the distance between adjacent predicted location points to the time interval to obtain the predicted movement speed corresponding to the predicted location point. Adjacent predicted location points refer to two consecutive predicted location points on the predicted route in the time series, representing the expected location status of the monitored object at consecutive moments. Predicted movement speed refers to the estimated movement speed of the monitored object in future time periods calculated based on the predicted route, used to assess movement status and behavioral patterns.

[0066] Specifically, the system calculates the movement speed of adjacent predicted locations along the predicted route. The calculation process first iterates through all predicted locations along the route in timestamp order, identifying adjacent point pairs. For each pair of adjacent locations, the system extracts the latitude and longitude coordinates and the corresponding timestamp data. Distance calculation uses a spherical distance algorithm, treating the Earth's surface as a sphere and calculating the great circle distance between the two points using latitude and longitude coordinates. The calculation process converts the latitude and longitude coordinates to radians and applies spherical trigonometric functions to calculate the spatial distance, obtaining a distance value in meters. Time interval calculation obtains the time difference by subtracting the timestamps, converting it to a time value in seconds. The predicted movement speed is obtained by dividing the distance value by the time interval value, with the result expressed in meters per second. To ensure calculation accuracy, the system detects and corrects abnormally large distance values ​​or time intervals, eliminating calculation deviations caused by positioning errors or time jumps. After the speed calculation is completed, each predicted movement speed value is associated with the corresponding predicted location point, establishing a mapping relationship between location points and speed attributes.

[0067] S403: For each predicted location point on the predicted route, query the preset environment database based on the coordinate information of the predicted location point to obtain the predicted environment type corresponding to the predicted location point; An environmental database refers to a pre-built data storage system containing environmental classification information for various geographical locations. It categorizes and manages the environmental attributes of different locations according to geographic coordinates. Predicted environmental types represent the classification of the geographic environment of a predicted location, including different environmental categories such as residential areas, commercial areas, schools, parks, roads, and rivers, reflecting the functional attributes and environmental characteristics of that location. Environmental classification is based on the land use classification standards and functional zoning principles of geographic information systems, assigning corresponding environmental type labels to location points.

[0068] Specifically, an environment type query operation is performed for each predicted location point on the predicted route. The system extracts the latitude and longitude coordinates of each predicted location point as query input and initiates the spatial query function of the environment database. The environment database adopts a hierarchical storage structure, storing different types of environmental information in corresponding data layers, including building layers, road network layers, water body layers, and green map layers. The query process determines the environmental attributes of the location point through multi-layer spatial overlay analysis and uses a point location query algorithm to detect the spatial relationship between the predicted location point and each environment layer. For location points belonging to multiple environment types, the system uses priority rules to determine the primary environment type, selecting according to the priority order of buildings, roads, water bodies, and green spaces. The query algorithm first checks whether the location point is located inside a building polygon; if it is, it further queries the functional attributes of the building to determine the specific environment type. For location points located on roads, buffer analysis is used to determine the distance relationship with the road centerline; location points closer to the road centerline are identified as road environment types.

[0069] S404: The regional risk level, predicted movement speed, and predicted environment type are considered as multiple influencing factors for the predicted location point.

[0070] Specifically, the three types of attribute information obtained in the aforementioned steps are integrated into a comprehensive set of influencing factors for each predicted location point. The integration process first establishes an attribute data structure for the predicted location points, creating a complete record for each point containing coordinate information, a timestamp, regional risk level, predicted movement speed, and predicted environment type. Data integration employs structured storage, organizing and associating various influencing factors according to a predetermined data format. The regional risk level, as a spatial security influencing factor, reflects the inherent risk level of the area where the location point is located; its value typically ranges from 1 to 3, with higher values ​​indicating higher risk. Predicted movement speed, as a dynamic behavior influencing factor, reflects the activity level and behavioral state of the monitored object; abnormally high or low speeds indicate potential risks. The predicted environment type, as an environmental background influencing factor, reflects the functional attributes and environmental characteristics of the location point; different environment types correspond to different security risk patterns and monitoring strategies. The integration of influencing factors also includes data standardization, converting factors with different dimensions and numerical ranges into a unified numerical interval to facilitate subsequent numerical calculations and comparative analysis. The system assigns a corresponding weight coefficient to each influencing factor, reflecting the importance of different factors to the overall risk assessment. The integrated influencing factor data is stored in a vectorized format, forming a multi-dimensional feature vector corresponding to each predicted location point. This influencing factor data will serve as input to the risk assessment model, used to calculate the overall risk level of the predicted route and identify high-risk sections.

[0071] Optionally, the final risk score can be revised in the following ways: Each influencing factor is compared with a preset regional correction table, a preset speed correction table, and a preset environmental correction table to obtain the corresponding correction value for each influencing factor; the initial risk score is then corrected based on each correction value to obtain the final risk score.

[0072] The regional correction table is a pre-defined lookup table that maps regional risk levels to risk correction values, containing numerical correction coefficients corresponding to different risk levels. The speed correction table is a lookup table that maps predicted movement speed ranges to corresponding correction values, used to determine the risk adjustment range based on the degree of anomaly in movement speed. The environmental correction table is a lookup table that maps different environmental types to corresponding correction values, reflecting the degree of impact of different environmental backgrounds on risk assessment.

[0073] Specifically, the system performs correction value lookup operations for each of the three influencing factors at each predicted location point. First, the system accesses the regional correction table, searching for the corresponding regional correction value based on the regional risk level of the predicted location point. The regional correction table uses a key-value pair storage structure, with the risk level as the lookup key and the correction value as the returned result. Low-risk areas correspond to negative correction values, used to reduce the base risk score; medium-risk areas correspond to zero correction values, keeping the base score unchanged; and high-risk areas correspond to positive correction values, used to increase the risk score. The speed correction value is obtained by comparing the predicted movement speed with the speed intervals in the speed correction table. The speed correction table divides movement speed into multiple intervals, each corresponding to a specific correction value. Normal speed ranges correspond to zero correction values, while abnormally low and abnormally high speeds both correspond to positive correction values, reflecting the amplifying effect of movement anomalies on risk. For speed values ​​located at interval boundaries, a linear interpolation method is used to calculate the accurate correction value. The environmental correction value is obtained through a direct mapping between the environmental type and the environmental correction table. The environmental correction table sets fixed correction values ​​for each environmental type. Densely populated areas such as commercial districts and schools correspond to higher positive correction values, while relatively safe areas such as parks and residential areas correspond to lower or negative correction values.

[0074] The correction calculation employs a weighted summation method, using the initial risk score as the base value and various correction values ​​as adjustment parameters for comprehensive calculation. The calculation process first determines the weight coefficients of each correction value: the regional correction value is typically weighted at 0.4, reflecting the significant impact of geographical location on risk; the speed correction value is weighted at 0.3, reflecting the risk contribution of abnormal movement behavior; and the environmental correction value is weighted at 0.3, reflecting the safety impact of the environmental background. The correction calculation involves multiplying each correction value by its corresponding weight, and then summing all weighted correction values ​​to obtain the total correction amount. The final risk score is obtained by adding the total correction amount to the initial risk score, achieving precise adjustment of the original score.

[0075] Based on the above embodiments, as an optional embodiment, in order to further reduce the power consumption of the anti-loss terminal in actual use and improve the battery life of the anti-loss terminal, the following steps S501-S505 can be specifically implemented: S501: Based on the historical movement trajectory data of the monitored object, identify trajectories that repeat more than a preset threshold as fixed travel routes; The preset frequency threshold refers to a numerical standard used to judge the frequency of trajectory repetition. When the number of times a trajectory is repeated exceeds this threshold, the system identifies it as a regular behavioral pattern. Fixed travel routes represent the frequently used movement paths of the monitored object, reflecting the monitored object's daily activity patterns and behavioral habits. The number of repetitions represents the statistical number of times a specific trajectory has been executed repeatedly in historical data, identified and counted through trajectory similarity comparison and pattern matching algorithms.

[0076] Specifically, the identification process first preprocesses historical trajectory data, including trajectory segmentation, noise filtering, and data standardization, to ensure the quality and consistency of the trajectory data. The system employs a trajectory clustering algorithm to analyze the similarity of all historical trajectories, grouping similar trajectories into the same category by calculating spatial distance and shape similarity. Trajectory similarity calculation is based on dynamic time warping and the longest common subsequence algorithm, comprehensively considering the spatial location, direction of movement, and temporal characteristics of the trajectories. The clustering process uses density clustering, automatically determining the number of clusters and identifying abnormal trajectories, avoiding the subjectivity of manually set parameters. For each trajectory cluster, the system counts the number of trajectories included as the repetition count, comparing it with a preset threshold. The preset threshold is typically set to 5 to 10 repetitions to ensure that the identified fixed routes have sufficient statistical significance. Trajectory clusters with repetition counts exceeding the threshold are identified as fixed travel routes, and the system generates representative trajectories for each fixed route as standard templates. Representative trajectories are generated by spatial averaging and path fitting of all trajectories within a cluster, preserving the main movement patterns and key location points.

[0077] S502: Construct a historical risk score curve for the fixed travel route based on the average historical risk score of each location point on the fixed travel route; Historical risk scores refer to risk assessment values ​​calculated for each location along a fixed travel route over a historical period, reflecting the historical safety status of different locations. The historical risk score curve is a continuous curve formed by connecting the average risk scores of each location along the fixed travel route in sequence, visually demonstrating the risk distribution characteristics of the entire route.

[0078] Specifically, the construction process first extracts representative trajectories of fixed travel routes as baseline paths, dividing the paths into equidistant sequence of location points at fixed intervals. The system traverses all historical trajectory instances corresponding to the fixed route, collecting risk score data from all historical passage records for each location point. The risk score data collection employs a spatial buffer method, establishing a small buffer around each location point and collecting risk scores from all historical location points within the buffer. For each location point, the system calculates the arithmetic mean of all collected risk scores to obtain the historical average risk score for that location. The average calculation process includes outlier detection and processing, identifying and removing or correcting risk scores that significantly deviate from the normal range. After calculation, the system arranges the average risk scores of each location point according to the spatial order of the route, forming a discrete risk score sequence. To construct a continuous risk score curve, the system uses a cubic spline interpolation algorithm to fit the discrete score points, generating a smooth and continuous curve function. The interpolation process maintains the original score point values ​​while ensuring a smooth and natural transition between adjacent points.

[0079] S503: Calculate the gradient of the historical risk score curve and determine the points where the gradient of the score change is greater than the preset gradient threshold as feedback monitoring nodes; The rating change gradient refers to the derivative of the historical risk rating curve at a specific location point, representing the rate and trend of risk rating change with location. The preset gradient threshold represents a numerical standard for judging the significance of risk changes; when the absolute value of the gradient at a location point exceeds this threshold, it indicates a significant risk change at that location. Feedback monitoring nodes refer to key locations along fixed travel routes where risk changes are drastic. These locations require enhanced monitoring and status feedback to promptly detect and respond to risk changes.

[0080] Specifically, the gradient calculation process first performs mathematical differentiation on the historical risk score curve, using numerical differentiation methods to calculate the derivative values ​​at each point on the curve. The numerical differentiation employs a central difference scheme, calculating approximate derivatives through the differences in function values ​​between adjacent points and the positional intervals. For the endpoints of the curve, forward or backward difference methods are used for gradient calculation. The system samples the entire fixed travel route at uniform intervals, calculating the corresponding gradient value at each sampling point to form a gradient distribution sequence. Gradient calculation also needs to consider the smoothness of the curve, eliminating noise interference during the calculation process through moving averages or low-pass filtering. The calculated gradient values ​​include positive and negative values; positive values ​​indicate that the risk score increases along the route, while negative values ​​indicate that the risk score decreases. The system calculates the absolute value of the gradient at each sampling point and compares it with a preset gradient threshold. The preset gradient threshold is adaptively set based on the numerical range of the risk score and the route length, ensuring that the number of identified monitoring nodes is appropriate and their distribution is reasonable. Points whose absolute gradient values ​​exceed a threshold are identified as candidate monitoring nodes. The system performs cluster analysis on these candidate nodes, merging adjacent nodes into a single monitoring node to prevent overly dense clustering. The final selected feedback monitoring nodes must meet minimum spacing requirements to ensure a reasonable spatial distribution among them.

[0081] S504: When the overlap between the predicted route and the fixed travel route is greater than the preset overlap threshold, it is determined that the monitored object is moving along the fixed travel route. Among them, overlap refers to the degree of spatial overlap between the predicted route and the fixed travel route. The similarity and consistency of the two routes are calculated by the path matching algorithm.

[0082] Specifically, the overlap calculation process first performs spatial alignment and standardization on the predicted route and the fixed travel route to ensure that both routes use the same coordinate system and sampling density. The system uses a dynamic time warping algorithm to optimally match the two routes, finding the correspondence between each point on the predicted route and the nearest point on the fixed route. The matching process comprehensively considers the consistency of spatial distance and direction of movement, calculating the matching quality through a weighted distance function. The overlap calculation is based on the statistical distribution of distance between matched point pairs. When the proportion of point pairs with matching distances less than a set threshold exceeds the overlap threshold, the two routes are considered to overlap. The distance threshold is typically set between 50 and 100 meters to ensure a reasonable tolerance range for the matching. The system also uses path length comparison as a supplementary indicator for overlap calculation. When the ratio of the predicted route length to the fixed route length is within a reasonable range, the reliability of the overlap determination is enhanced. The overlap calculation employs a parallel processing approach, simultaneously comparing with multiple fixed travel routes and selecting the route with the highest overlap as the matching result. Upon successful matching, the system records the correspondence between the predicted route and the fixed route, including the identifier of the matched fixed route, the overlap value, and the matching confidence level. The judgment result triggers the corresponding control logic, switching the monitoring mode to fixed route monitoring mode and enabling the corresponding feedback monitoring node configuration.

[0083] S505: Controls the anti-loss terminal to report the actual status signal to the LoRa gateway only when it reaches the feedback monitoring node, and reduces the reporting frequency when it is not at the feedback monitoring node.

[0084] The reporting frequency refers to the time interval and number of times the anti-loss terminal sends status signals to the LoRa gateway. The frequency setting directly affects battery consumption and communication overhead. The actual status signal contains key information such as the anti-loss terminal's current location, timestamp, and battery status, used for real-time monitoring and trajectory tracking.

[0085] Specifically, the control process first establishes a location database for feedback monitoring nodes in the anti-loss terminal, storing the geographic coordinates and monitoring parameters of all monitoring nodes. The terminal obtains its current location information in real time via GPS module and performs location matching with the feedback monitoring node database. Location matching uses a distance threshold method; when the distance between the terminal's location and a certain monitoring node is less than a preset threshold, the terminal is determined to have reached that monitoring node. The distance threshold is typically set to 20 to 50 meters to ensure the accuracy and timeliness of location determination. When the terminal reaches a feedback monitoring node, the system immediately triggers a status signal reporting operation, sending a complete status signal containing current location, timestamp, and movement status information to the gateway via the LoRa communication module. The reporting operation is processed with high priority to ensure that the status information of the monitoring node can be transmitted to the monitoring center in a timely manner. At non-feedback monitoring node locations, the terminal sends status signals at a reduced reporting frequency. The specific methods for reducing the frequency include extending the reporting interval and reducing the data packet size. The normal reporting interval is usually 1 to 5 minutes, and after reducing the frequency, the interval is extended to 10 to 30 minutes. The system also employs an intelligent reporting strategy, maintaining a relatively high reporting frequency while in motion and further reducing the frequency when stationary. The frequency control mechanism includes anomaly handling; when an emergency or abnormal movement is detected, the system automatically resumes high-frequency reporting.

[0086] The following are system embodiments of this application, which can be used to execute the method embodiments of this application. For details not disclosed in the system embodiments of this application, please refer to the method embodiments of the application.

[0087] Please see Figure 2 This illustration shows a schematic diagram of a special population monitoring system based on the LoRa wireless communication protocol provided in an exemplary embodiment of this application. The system can be implemented as all or part of a larger system through software, hardware, or a combination of both. The special population monitoring system based on the LoRa wireless communication protocol includes: The initial scoring module is used to receive the current status signal sent by the anti-loss terminal through the LoRa gateway, and calculate the initial risk score representing the current risk level based on the current status signal. The prediction module is used to input the current status signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and to determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The correction module is used to correct the initial risk score based on the influencing factors to obtain the final risk score; The monitoring module is used to determine the target monitoring mode of the anti-loss terminal based on the score range of the final risk score, so that the anti-loss terminal can monitor the monitored object according to the target monitoring mode.

[0088] This application also provides a computer storage medium that can store multiple instructions. The instructions are adapted to be loaded and executed by a processor as described in the above embodiments for monitoring special populations based on the LoRa wireless communication protocol. For details of the execution process, please refer to the specific description of the embodiments, which will not be repeated here.

[0089] Please see Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 300 may include: at least one processor 301, at least one network interface 304, user interface 303, memory 305, and at least one communication bus 302.

[0090] The communication bus 302 is used to enable communication between these components.

[0091] The user interface 303 may include a display screen and a camera.

[0092] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0093] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling data stored in the memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of digital signal processing, field-programmable gate array, or programmable logic array. The processor 301 may integrate one or more of the following: a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0094] The memory 305 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 305 may include a non-transitory computer-readable medium. The memory 305 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 3 As shown, the memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a special population monitoring method based on the LoRa wireless communication protocol.

[0095] exist Figure 3 In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 301 can be used to call an application stored in the memory 305 for a special population monitoring method based on the LoRa wireless communication protocol. When executed by one or more processors, the electronic device executes one or more methods as described in the above embodiments.

[0096] An electronic device readable storage medium stores instructions that, when executed by one or more processors, cause the electronic device to perform one or more methods as described in the above embodiments.

[0097] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0098] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0100] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0101] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0102] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0103] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truths. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure.

Claims

1. A method for monitoring special populations based on the LoRa wireless communication protocol, characterized in that, The method includes: The LoRa gateway receives the current status signal sent by the anti-loss terminal and calculates an initial risk score representing the current risk level based on the current status signal. The current status signal is input into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and to determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The initial risk score is revised based on the influencing factors to obtain the final risk score; Based on the score range of the final risk score, the target monitoring mode of the anti-loss terminal is determined so that the anti-loss terminal monitors the monitored object according to the target monitoring mode.

2. The method according to claim 1, characterized in that, The current status signal includes the location information, motion status information, and environmental information of the anti-loss terminal. The calculation of an initial risk score characterizing the current risk level based on the current status signal includes: Based on the location information, the safe distance between the current location of the anti-loss terminal and the boundary of the preset safe area is determined. The safe distance is compared with multiple preset distance thresholds, and the score corresponding to the distance interval to which the safe distance belongs is determined as the location risk score. Based on the motion state information, the moving speed of the anti-loss terminal is determined, the moving speed is compared with multiple preset speed thresholds, and the score corresponding to the speed range to which the moving speed belongs is determined as the motion risk score. Based on the environmental information, the environmental type of the anti-loss terminal is determined, and the corresponding risk coefficient is found in the preset environmental type-risk coefficient mapping table according to the environmental type. The risk coefficient is then determined as the environmental risk score. The location risk score, the movement risk score, and the environmental risk score are weighted and summed according to preset weighting coefficients to obtain the initial risk score.

3. The method according to claim 1, characterized in that, Before inputting the current state signal into the historical behavior trajectory model, the method further includes: The monitored object acquires multiple historical movement trajectory data within a historical time period. Each historical movement trajectory data includes multiple historical location points arranged in chronological order and a timestamp corresponding to each historical location point. For each historical movement trajectory data, feature extraction is performed to obtain a trajectory feature vector, which includes movement direction feature, movement speed feature, dwell time feature, and trajectory repeatability feature; The trajectory feature vectors corresponding to the multiple historical movement trajectory data are used as training samples, and the training samples are trained using a sequence prediction model to obtain the historical behavior trajectory model.

4. The method according to claim 1, characterized in that, The step of inputting the current state signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period includes: The current state signal is input into the historical behavior trajectory model to extract the current trajectory feature vector, which includes the current movement direction feature, the current movement speed feature, and the current position feature. Based on the current trajectory feature vector, sequence prediction is performed using the historical behavior trajectory model to generate multiple predicted location points within the preset time period. The multiple predicted location points are then connected in chronological order to form the predicted route.

5. The method according to claim 1, characterized in that, The determination of multiple influencing factors corresponding to each location point on the predicted route includes: For each predicted location point on the predicted route, a preset regional attribute database is queried based on the coordinate information of the predicted location point to obtain the regional risk level corresponding to each predicted location point. For adjacent predicted location points on the predicted route, the ratio of the distance between adjacent predicted location points to the time interval is calculated to obtain the predicted movement speed corresponding to the predicted location point. For each predicted location point on the predicted route, a preset environment database is queried based on the coordinate information of the predicted location point to obtain the predicted environment type corresponding to the predicted location point; The regional risk level, the predicted movement speed, and the predicted environment type are considered as multiple influencing factors corresponding to the predicted location point.

6. The method according to claim 5, characterized in that, The step of revising the initial risk score based on the influencing factors to obtain the final risk score includes: Each of the aforementioned influencing factors is compared with a preset regional correction table, a preset speed correction table, and a preset environmental correction table to obtain the correction value corresponding to each of the aforementioned influencing factors; The initial risk score is corrected based on the aforementioned correction values ​​to obtain the final risk score.

7. The method according to claim 1, characterized in that, The method further includes: Based on the historical movement trajectory data of the monitored object, trajectories that repeat more than a preset threshold number of times are identified as fixed travel routes. Based on the average historical risk scores of each location point on the fixed travel route, a historical risk score curve corresponding to the fixed travel route is constructed. Calculate the gradient of the score change in the historical risk score curve, and determine the points where the gradient of the score change is greater than a preset gradient threshold as feedback monitoring nodes; When the overlap between the predicted route and the fixed travel route is greater than a preset overlap threshold, it is determined that the monitored object is moving along the fixed travel route. The anti-loss terminal is controlled to report the actual status signal to the LoRa gateway only when it reaches the feedback monitoring node, and the reporting frequency is reduced when it is not at the feedback monitoring node.

8. A special population monitoring system based on the LoRa wireless communication protocol, characterized in that, The system includes: The initial scoring module is used to receive the current status signal sent by the anti-loss terminal through the LoRa gateway, and calculate an initial risk score representing the current risk level based on the current status signal. The prediction module is used to input the current status signal into the historical behavior trajectory model to obtain the predicted route of the monitored object to which the anti-loss terminal belongs within a preset time period, and to determine multiple influencing factors corresponding to each location point on the predicted route. The historical behavior trajectory model is trained based on the historical movement trajectory data of the monitored object. The correction module is used to correct the initial risk score based on the influencing factors to obtain the final risk score; The monitoring module is used to determine the target monitoring mode of the anti-loss terminal based on the score range of the final risk score, so that the anti-loss terminal monitors the monitored object according to the target monitoring mode.

9. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions, which are adapted to be loaded by a processor and executed as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, The device includes a processor, a memory, and a transceiver, wherein the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 7.

Images